Squirrels vs Motorways

Red Squirrels (Sciurus vulgaris) have been in decline as a staple part of British wildlife since the introduction of their grey counterparts (Sciurus carolinensis) as an invasive species.

Here, an assessment of the balance of squirrel colours is presented, and a case made for how habitat fragmentation can help protect native wildlife from invasive species.

It uses data from the NBN Atlas, which is a voluntary dataset of species sightings, and landuse data from Digimap .

Fig 1. Ratio of Red to Grey Squirrels by UK County Fig 2. UK Habitat Fragmentation: Woodland & Motorways

There is an unsurprising dominance of grey squirrels in England, but there are pockets where red squirrels are thriving. Red squirrels are dominant in Scotland and most of Northern Ireland.

Figure 2 can explain some of the reasons behind the findings of figure 1 by showing the UK’s coniferous woodland and the state of its discontinuity, as well as the motorway network as contributors to habitat fragmentation.

It shows how red squirrels are most frequently found in isolated areas, with three types of boundary protecting their territory:

  1. Water: The Isle of Wight and Brownsea Island are excellent examples of how large bodies of water can isolate a population and keep competitors out.

  2. Roads: Major roads isolate communities and keep red and grey populations from mixing.

  3. Low woodland density: Figure 2 shows a large area across the Scottish borders with a relatively low density of woodland, where numbers of both red and grey squirrels are very low. This habitat gap insulates the red squirrels in Scotland from invading grey squirrels from the south. The population here is also split north/ south by the M9 motorway and the land is split in the east by the River Forth, making it an excellent example of all three factors, as is Southport. Southport is bordered by the Irish Sea to the west, has very isolated coniferous woodland, and its red squirrels are only found west of the A595.

A Quick Statistical Analysis

Hypothesis: Areas with high numbers of red squirrels will have woodland that is statistically more fragmented.

The UK was split into polygons isolated by major roads (a more meaningful division than UK counties), to which a fragmentation equation was applied to calculate a value for how fragmented the woodland was i.e. is the woodland one whole unit of 100 or 100 fragments of 1 ?

Figure 1

Road Polygons.png

There is a slight negative correlation between woodland fragmentation and squirrel density, indicating that red squirrels prefer more connected habitat, which goes against the hypothesis made. The R2 value, however, is very low and the covariance is very wide, indicating a lack of correlation between the two. The red squirrel density compared with the raw number of woodlands per area shows a stronger and tighter negative correlation, suggesting that there are more red squirrels in areas with fewer individual woodlands. This supports the hypothesis, but may indicate that the equation used to calculate the fragmentation was less good a measure of habitat fragmentation than the simple number of woodlands. The R2 value is still low, though, at -0.004.

fig5.png fig6.png
R2=0.052, X-Var=48170.585, Y-Var=0.031, CoVar=1.918 R2=-0.004, X-Var=0.031, Y-Var=66254.798, CoVar=-0.162_

An independent T-Test was conducted to test whether there is a statistically significant difference between the values in the highest and lowest quartiles for fragmentation and woodland counts.

Fragmentation Raw woodland count
P value 0.04 0.02
T value -2.04 -2.31

This suggests that there may be a statistical difference, but it is so close to the critical value of 0.05 that the null hypothesis cannot be rejected on this basis.

However, the low P value for woodland count shows that there may be a relationship between the number of individual woodlands and the number of squirrels. Considering this in context with the low R2 values though means that there is not enough evidence to fully reject the null hypothesis, and a statistical link cannot therefore be made.

Limitations:

  • Sightings are from a voluntary dataset, and may therefore be unreliable. It is unknown how much data is missing, wasn’t recorded, or was recorded incorrectly.
  • Absence of data in a particular area does not necessarily mean that the squirrels were not there, but that the data simply wasn’t recorded.
  • The small sample size of red squirrel populations is a limiting factor. Future research would benefit from testing this approach on species with more numerous data.

References